In this study we build a sentiment analysis system. We based our system upon the Naïve Bayes classifier.
First we applied two text pre-processing steps. (1) Remove punctuation and stop-words and (2) take the stems of the words.
After that we implemented Naive-Bayes algorithm without third party libraries.
We evaluate the performance of our model by our self-written F1-score function.
In the last section we give the results of the model with and without text pre-processing techniques.
According to our results using both techniques improves the F1-score.